Abstract
Abstract Background Microglia are the innate immune cells in the central nervous system that responds to brain injury and pathogens. The loss of homeostatic microglia is associated with neurodegenerative diseases. Multiple studies of AD mouse models have identified AD‐associated microglia subtypes, e.g., disease associated microglia (DAM). However, studies of human microglia have not reported similar microglia subtypes. Therefore, how human microglia change in the presence of amyloid‐ protein deposition and phosphorylated Tau remains an open question. Method We have developed an approach that estimates transcriptional kinetics from scRNA‐seq data. This approach models how “bursty” and variable a gene expression is. We hypothesize that transcriptional bursts of key genes are different within AD cases compared to MCI individuals. We used the ROSMAP scRNAseq microglia dataset of 8 AD cases and 4 MCI cases and applied our approach to estimate the transcriptional kinetics for each subject. We then compared the difference in transcriptional kinetics of genes that are estimated both in AD and MCI. Result First, we found that 96% of microglia cells in AD and MCI shows a similar co‐occurrence pattern of expressed genes and, therefore, likely reflect the same cell type. Within the assumed homogeneous microglia cells, we identified two genes, C1QC and ACTB, showing significant differences in their transcriptional kinetics between AD and MCI cases. Both C1QC and ACTB show larger burst sizes in MCI than in AD cases (p‐value for C1QC:0.0038; p‐value for ACTB: 0.00918). Consequently, C1QC and ACTB show a higher expression level variability among microglia cells in MCI cases than AD. Conclusion Our result suggests C1QC and ACTB may be associated with AD risk via changes in variability of gene expression rather than total gene expression levels. The increased variability in transcription of C1QC and ACTB in MCI individuals relative to AD cases might indicate more mobile and active physiological states of microglia. These results may point to functional changes in these genes with AD progression and are prime targets for further experimental validation.